We combine growth theory with US Census data on individual schooling and wages to estimate the aggregate return to human capital and human capital externalities in cities. Our estimates imply that a one-year increase in average schooling in cities increases their aggregate labor productivity by 8 to 11 percent. We find no evidence for aggregate human capital externalities in cities however although we use three different approaches. Our main theoretical contribution is to show how human capital externalities can be identified (non-parametically) even if workers with different levels of human capital are imperfect substitutes in production.